CrtxReavr: come on ZFS.....

yeah - I know it is stupid --- I had ZFS raid before - it was good - but to have ZFS on a single drive is plain ignorance

ZFS on a single drive is like tits on a bull.

lol

I know

Not sure you do.

I think I do now, but if you care to share more knowledge, I am all eyes

The point of ZFS (and most RAID) is redundancy. Using it on a single drive (spindle) just slows your machine down and provides zero upside.

It'd say it's like using RAID0, but it's actually worse than that, because at least RAID0 speeds up reads and writes.

Not true. You still get snapshots and ZFS send/recv. :)

Right up until that single drive dies.

clones, boot environments, checkpoints, dedup... plenty of reasons to use zfs on a single-drive system, like a laptop

That's why you send/recv.

Truths of Information Technology: Storage fails.

using zfs on a single drive is no worse than using ufs on a single drive

Nope. And I'd take ZFS over UFS any day of the week.

'Cept UFS is a lot faster.

Then you're ignorant.

Because I'd take resiliency over speed?

i feel like most people running on a single disk are not running performance critical workloads, and probably care a lot more about features like boot environments and cheap jails

so saying it's "ignorant" to use zfs instead of ufs in that situation seems pretty silly

Or... ignorant?

I run compression on my single ZFS boot drive.

Nothing wrong with that.

I've got a few other drives in mirrors and such, but having ZFS as the sole manager of my disks keeps me from being an accidental moron.

Safety nets are handy to have.

Danger nets, however, are not.

(also, single disk is basically the only situation where copies=2 is useful, except with SSDs it's probably still not useful as they don't tend to fail that way)

I'll be building an 8-drive box here "soon"

Think I'm going to have a pool of two disk mirrors

I need another 8TB or 2, as I still have *cough* ntfs disks I need to convert

Slap one o' them in your drive bay. No need to worry then X-D

oh goodness, that could cover my issue

Refurbs are nice.

bookmarking for future reference :) thanks

for some reason i find the idea of a 26TB 512e drive slightly hilarious... isn't everyone using 4k-aware systems now?

(i guess you can just reformat it to 4kn though)

I don't know what that even is :O

the sector size default on the drive

some of the lower capacity ones are 4Kn

thedaemon: 512e means it uses 4K sectors but pretends to the OS that it uses 512-byte sectors (which is why the zfs ashift option exists...)

The drive has sectors in 512-bytes vs 4kb

thanks, gotcha.

hey SponiX

thedaemon: filed my 1st FreeBSD-Current bug report today --> bugs.freebsd.org/bugzilla/show_bug.cgi?id=288084

:) cool

SponiX: that's already fixed, update and rebuild kernel

oh, i see kib already commented

ivy: I often buy the 512 as sometimes they are significantly cheaper LOL

ivy: It is actually a kernel bug, and not a tail bug?

SponiX: yes, it's a kqueue bug

I've done a system update since then. Just haven't rebooted yet

Okay, I will reboot and see if it goes away

SponiX: cgit.freebsd.org/src/commit/?id=13a…c0c0e3db6e2c1a317031c7d1bf4651bec3d

I want points for effort anyway damn it :P

<-- rebooting

well, i appreciate you filing bugs :-)

at least this one wasn't my fault

That's always nice.

I'm wondering if the bug was a typo or if the code was re-written and the test was actually expected to be true.

I'll have to dig into when I have the time just out of curiosity.

ivy: Yeah, it is fixed already. That was FAST

SponiX: unsurprisingly a lot of people immediately complained on the mailing list when tail -f stopped working :-)

Yeah, looking at logs is kinda important in the Unix world - who would have thought LOL

I kept looking at my unbound configuration thinking it was an unbound problem for a bit.

well, I gotta get to work now

I am still confused about ZFS copy on write feature. Is it enabled by default ? How do you go back on file level to saved previous version etc ?

acu: copy-on-write is used automatically for snapshots and clones, and it's also support by commands like cp(1) (this is a newer feature called block cloning). usually it's not something you need to worry about - if you want to be able to roll back to previous versions of files, use snapshots

ivy: I thought COW is anytime you open a file like myschedule.odt it will automaically create a copy on it.. so you will actually have a new copy of the file while the old one is "hidden" - maybe my knowledge is not up to par on this - but the name suggest - any time you write something on a file - you make a copy of it...

no, that's called "file versioning" (as seen in operating systems like OpenVMS) and ZFS doesn't support it. copy-on-write means the filesystem internally avoids copying blocks when doing things like creating snapshots

it's true that writing to a file on zfs writes new data instead of overwriting the existing blocks, but that functionality is not exposed to the user directly, and it's on the block level, not the file level

ivy thanks - I think you should make a youtube about it --- zfs is not trivial..

i'm fairly sure many people have already made youtube videos about zfs

(starting with the Sun one about yelling at your disk array)

there are a lot - all confusing...

you seem to really know your stuff... I mean it... it should be a practical usage both on freebsd and debian - all what you said...real life examples